<div dir="ltr">Hi Sergey,<br><br>so the problem got fixed after I turned off WITH_CYCLES_DEBUG from CMake.<br>to clarify what happens when WITH_CYCLES_DEBUG is active:<br><br>scene->mutex._Mtx points at the correct memory.<br>once you enter the function, _Mtx points at a bad memory (16 bytes offset).<br><br>this is caused by DeviceScene dscene;<br><br>if you add something like:<br>int anything[4000];<br>thread_mutex mutex;<br>int something[4000];<br><br>it works..., also SceneParams params; memory is dumbed, so you may check what cycles debug is doing with memory (a dangling pointer is changing memory).<br><br>thanks again for help.<br><br>cheers,<div>Mohamed Sakr</div></div><div class="gmail_extra"><br><div class="gmail_quote">On Mon, Aug 29, 2016 at 9:45 AM, Sergey Sharybin <span dir="ltr"><<a href="mailto:sergey.vfx@gmail.com" target="_blank">sergey.vfx@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr">Hi,<div><br></div><div>Several points:</div><div><br></div><div>1. Such errors are really difficult to give any meaningful suggestions unless they are reproducable with official builds (like Blender). There are lots of possibilities here starting from real issue in the code ending with wrongly compiled dependencies.</div><div><br></div><div>2. There are fixes which were synchronized to standalone repo this morning which solves some known crashes in subd code.</div><div><br></div><div>3. If that's recently introduced error you can easily isect the sources and find out a buggy commit, so we can have a closer look on it.</div></div><div class="gmail_extra"><br><div class="gmail_quote"><div><div class="h5">On Mon, Aug 29, 2016 at 9:35 AM, Mohamed Sakr <span dir="ltr"><<a href="mailto:3dsakr@gmail.com" target="_blank">3dsakr@gmail.com</a>></span> wrote:<br></div></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div><div class="h5"><div dir="ltr">Hi,<br><br>I just grabbed the most recent build (18/8), I'm on Windows 10, VS2013, C++11<br>there is a decent bug with the scene->mutex. (actually the bug is caused by the member DeviceScene dscene;)<br><br>old build was running fine, new build got a problem which I'll explain<br><br>what happens:<br>memory is fine, so at any point in my DLL, if I check scene->mutex, memory is correct.<br>now the problem comes when I try to access it:<br><br>if (_session && _session->scene && _session->ready_to_reset() && _session->scene->mutex.try_loc<wbr>k())<br><br>here, scene->mutex.try_lock(), inside the function call, the mutex pointer is WRONG.<br><br>after debugging, pointer is shifted by 2 indices (16 bytes)<br><br>so outside the function, (scene->mutex._Mtx) is correct<br>inside the function, *(&_Mtx + 2) is correct<br><br>_Mtx alone is a corrupted memory, showing 0x0000...1<br><br>any ideas?<br><br>cheers,<br>Mohamed Sakr</div>
<br></div></div>______________________________<wbr>_________________<br>
Bf-cycles mailing list<br>
<a href="mailto:Bf-cycles@blender.org" target="_blank">Bf-cycles@blender.org</a><br>
<a href="https://lists.blender.org/mailman/listinfo/bf-cycles" rel="noreferrer" target="_blank">https://lists.blender.org/mail<wbr>man/listinfo/bf-cycles</a><br>
<br></blockquote></div><span class="HOEnZb"><font color="#888888"><br><br clear="all"><div><br></div>-- <br><div data-smartmail="gmail_signature"><div><span style="color:rgb(102,102,102)">With best regards, Sergey Sharybin</span></div></div>
</font></span></div>
<br>______________________________<wbr>_________________<br>
Bf-cycles mailing list<br>
<a href="mailto:Bf-cycles@blender.org">Bf-cycles@blender.org</a><br>
<a href="https://lists.blender.org/mailman/listinfo/bf-cycles" rel="noreferrer" target="_blank">https://lists.blender.org/<wbr>mailman/listinfo/bf-cycles</a><br>
<br></blockquote></div><br></div>